Manifold: A Model-Agnostic Visual Debugging Tool for Machine Learning at Uber | Uber Engineering Blog
Machine learning (ML) is widely used across the Uber platform to support intelligent decision making and forecasting for features such as ETA prediction and fraud detection. For optimal results, we invest a lot of resources in developing accurate predictive ML models. In fact, it’s typical for practitioners to devote 20 percent of their effort into building initial working models, and 80 percent of their effort improving model performance in what is known as the 20/80 split rule of ML model development.

Traditionally, when data scientists develop models, they evaluate each model candidate using summary scores like log loss, area under curve (AUC), and mean absolute error (MAE). Although these metrics offer insights into how a model is performing, they do not convey much information regarding why a model is not performing well, and from there, how to improve its performance. As such, model builders tend to rely on trial and error when determining how to improve their models.

To make the model iteration process more informed and actionable, we developed Manifold, Uber’s in-house model-agnostic visualization tool for ML performance diagnosis and model debugging. Taking advantage of visual analytics techniques, Manifold allows ML practitioners to look beyond overall summary metrics to detect which
machine-learning  visualization 
yesterday
Biomimetics | Free Full-Text | Biomechanics in Soft Mechanical Sensing: From Natural Case Studies to the Artificial World
Living beings use mechanical interaction with the environment to gather essential cues for implementing necessary movements and actions. This process is mediated by biomechanics, primarily of the sensory structures, meaning that, at first, mechanical stimuli are morphologically computed. In the present paper, we select and review cases of specialized sensory organs for mechanical sensing—from both the animal and plant kingdoms—that distribute their intelligence in both structure and materials. A focus is set on biomechanical aspects, such as morphology and material characteristics of the selected sensory organs, and on how their sensing function is affected by them in natural environments. In this route, examples of artificial sensors that implement these principles are provided, and/or ways in which they can be translated artificially are suggested. Following a biomimetic approach, our aim is to make a step towards creating a toolbox with general tailoring principles, based on mechanical aspects tuned repeatedly in nature, such as orientation, shape, distribution, materials, and micromechanics. These should be used for a future methodical design of novel soft sensing systems for soft robotics. View Full-Text
biomimicry  sensors 
5 days ago
Evolving embodied intelligence from materials to machines | Nature Machine Intelligence
Natural lifeforms specialize to their environmental niches across many levels, from low-level features such as DNA and proteins, through to higher-level artefacts including eyes, limbs and overarching body plans. We propose ‘multi-level evolution’, a bottom-up automatic process that designs robots across multiple levels and niches them to tasks and environmental conditions. Multi-level evolution concurrently explores constituent molecular and material building blocks, as well as their possible assemblies into specialized morphological and sensorimotor configurations. Multi-level evolution provides a route to fully harness a recent explosion in available candidate materials and ongoing advances in rapid manufacturing processes. We outline a feasible architecture that realizes this vision, highlight the main roadblocks and how they may be overcome, and show robotic applications to which multi-level evolution is particularly suited. By forming a research agenda to stimulate discussion between researchers in related fields, we hope to inspire the pursuit of multi-level robotic design all the way from material to machine.
artificial-intelligence  machine-learning  robotics 
5 days ago
Comet.ml cheat sheet: supercharge your machine learning experiment management
Comet.ml allows you to automatically track your machine learning code, experiments, hyperparameters, and results to achieve reproducibility, transparency, and more efficient iteration cycles. We…
machine-learning  logging 
4 weeks ago
NASA Goddard Workshop on Artificial Intelligence
Astrophysics researcg=h to support astroparticle (gamma-ray and cosmic-ray), x-ray, gravitational-wave, observational cosmology, exoplanet and stellar astrophysics.
machine-learning  nasa 
4 weeks ago
tensorlayer/tensorlayer: Deep Learning and Reinforcement Learning Library for Developers and Scientists
Deep Learning and Reinforcement Learning Library for Developers and Scientists - tensorlayer/tensorlayer
machine-learning 
5 weeks ago
The Microsoft Simple Encrypted Arithmetic Library goes open source - Microsoft Research
Homomorphic Encryption Homomorphic Encryption (HE) refers to a special type of encryption technique that allows for computations to be done on encrypted data, without requiring access to a secret (decryption) key. The results of the computations are encrypted, and can be revealed only by the owner of the secret key.
encryption  machine-learning 
6 weeks ago
Symposium on Blockchain for Robotic Systems — MIT Media Lab
A symposium bringing together key stakeholders to discuss the possibilities of blockchain technology in robotics.
blockchain  robotics 
6 weeks ago
« earlier      
3d 3d-printing 3d-scanning accessibility afrl ajax algorithms analytics android api application architecture arduino aria artificial-intelligence augmented-reality awesome benchmark big-data biomimicry biomimicry_petal biomimicry_petal_angeera biomimicry_petal_calvin biomimicry_petal_colleen biomimicry_petal_herb biomimicry_petal_nick biomimicry_petal_victoria biomimicry_petal_vik blog book books browser camera career cave cfd cheatsheet cleveland cloud cloudcomputing code collaboration color comparison computer computing course css d3 data data-cleanup data-management data-science data-visualization debugging deep-learning design development docker editor education educational example examples filetype:pdf flash forms fortran framework free gallery git google government gpu graphics grcsmac grcsocnet gui guide guidelines haptics hdf5 head-mounted-displays health high high-performance-computing history hololens html html5 image image-processing images immersive infographic infographics interactive interfaces iot ipython javascript jquery jupyter jupyter-notebook kinect leap leap-motion learning libraries library linux mac machine-learning management media:document medical medicine memory microsoft mixed-reality mobile modeling motion mpi nasa nasagrcsocnet natural natural-user-interfaces neural-network notebook oculus oculus-rift online open-source opensource optimization overview pandas parallel pdf performance ping.fm plotting policy presentations privacy processing profiling programming provenance pycon python r raspberry-pi reality reference reproducible reproducible-research research resources responsive rift robotics science scientific scientific-computing scientific-data-management scratch scripting search secondlife security selenium seo socialmedia socialnetworking software standards statistics telepresence tensorflow testing tips tool tools toread towatch tracking training tutorial tutorials twitter unity usability user validation video videos virtual virtual-reality virtualworlds visualization vive watched watson web web2.0 webdesign webdev webgl webvr windows workflow

Copy this bookmark:



description:


tags: